4–9 Oct 2026
Europe/Dublin timezone

Running a Science Cloud without being a Cloud Expert

6 Oct 2026, 16:00
1h 30m
Poster DASH General Poster Session

Speaker

Peter Shumate (Johns Hopkins University Applied Physics Lab)

Description

Cloud computing is an increasingly important platform for scientific collaboration, interactive analysis, and reproducible research. However, deploying and maintaining cloud infrastructure often requires familiarity with a rapidly growing ecosystem of tools, creating a significant barrier for many scientists and research software engineers. A reproducible deployment makes it practical to rapidly create new science environments for collaborations, workshops, or project-specific efforts, while leveraging cloud bursting to temporarily scale compute resources for computationally intensive analyses without maintaining permanently provisioned infrastructure. HelioCloud was designed to reduce this complexity by providing a reproducible, open-source science cloud built on modern cloud-native technologies (OpenTofu, Helm, Kubernetes) while exposing only the concepts necessary for day-to-day operation. We discuss how researchers and developers can confidently contribute to and operate a science cloud without becoming cloud experts, while taking advantage of scalable computing, simplified deployments, and direct access to large scientific datasets.

Author

Peter Shumate (Johns Hopkins University Applied Physics Lab)

Co-authors

Jeffrey Bradford (NASA) Omar Shalaby (NASA) Sarah Rourke (NASA) Brian Thomas (NASA) Lisa Knowles (Johns Hopkins University Applied Physics Lab) Nicholas Lenzi (Johns Hopkins University Applied Physics Lab) Jon Vandegriff (Johns Hopkins University Applied Physics Lab) Alex Antunes (Johns Hopkins University Applied Physics Lab)

Presentation materials

There are no materials yet.